bioRxiv preprint first posted online Feb. 6, 2017; doi: http://dx.doi.org/10.1101/106237. The copyright holder for this preprint (which was not
peer-reviewed) is the author/funder. It is made available under a CC-BY 4.0 International license.
Title: Education and coronary heart disease: a Mendelian
randomization study
Authors: Dr Taavi Tillmann1*†, Dr Julien Vaucher2*†, Ms Aysu Okbay3, Dr Hynek Pikhart1, Dr
Anne Peasey1, Dr Ruzena Kubinova4, Prof Andrzej Pajak5, Prof Abdonas Tamosiunas6, Prof
Sofia Malyutina7, 8, Dr Krista Fischer9, Dr Giovanni Veronesi10, Dr Tom Palmer11, Dr Jack
Bowden12, Prof George Davey Smith12, Prof Martin Bobak1, Dr Michael V Holmes13, 14
†=contributed equally
Affiliations:
1
2
3
4
Department of Epidemiology & Public Health, University College London, UK
Department of Internal Medicine, Lausanne University Hospital, Switzerland
Department of Applied Economics, Erasmus University, Netherlands
Department of Environmental Health Monitoring System, National Institute of Public Health, Prague, Czech
Republic
5
Department of Epidemiology and Population Studies, Institute of Public Health, Jagiellonian University,
Krakow, Poland
6
7
8
9
Institute of Cardiology, Lithuanian University of Health Sciences, Kaunas, Lithuania
Institute of Internal and Preventive Medicine, Novosibirsk, Russia
Novosibirsk State Medical University, Novosibirsk, Russia
Estonian Genome Center, University of Tartu, Estonia
10
11
12
13
Research Center in Epidemiology and Preventive Medicine, University of Insubria, Varese, Italy
Department of Mathematics and Statistics, Lancaster University, Lancaster, UK
Medical Research Council Integrative Epidemiology Unit at the University of Bristol, UK
Clinical Trial Service Unit & Epidemiological Studies Unit, Nuffield Department of Population Health,
University of Oxford, UK
14
Medical Research Council Population Health Research Unit at the University of Oxford, UK
*Correspondence to: [email protected] Department of Epidemiology & Public Health,
University College London, 1-19 Torrington Place, WC1E 7HB, UK.
And Julien Vaucher: [email protected], Service of Internal Medicine, Lausanne
University Hospital, Bugnon 46, 1011 Lausanne, Switzerland.
Abstract word count:
257
Word count (excl. references, figures & acknowledgements):
2433
Number of figures & tables:
5
Number of references:
34
1
bioRxiv preprint first posted online Feb. 6, 2017; doi: http://dx.doi.org/10.1101/106237. The copyright holder for this preprint (which was not
peer-reviewed) is the author/funder. It is made available under a CC-BY 4.0 International license.
Abstract:
Objectives: To determine whether educational attainment is a causal risk factor in the
development of coronary heart disease.
Design: Mendelian randomization study, where genetic data are used as proxies for
education, in order to minimize confounding. A two-sample design was applied, where
summary level genetic data was analysed from two publically available consortia.
Setting: In the main analysis, we analysed genetic data from two large consortia
(CARDIoGRAM and SSGAC), comprising of 112 cohorts from predominantly high-income
countries. In addition, we also analysed genetic data from 7 additional large consortia, in
order to identify putative causal mediators.
Participants: The main analysis was of 589 377 men and women, predominantly of
European origin.
Exposure: A one standard deviation increase in the genetic predisposition towards higher
education (i.e. 3.6 years of additional schooling). This was measured by 162 genetic variants
that have been previously associated with education.
Main outcome: Combined fatal and nonfatal coronary heart disease (63 746 events).
Results: 3.6 years of additional education lowered the risk of coronary heart disease by a
third (odds ratio = 0.67, 95% confidence interval [CI], 0.59 to 0.77, p=0.01). Equivalent
increases in education were also causally associated with reductions in smoking, BMI and
improvements in blood lipid profiles.
Conclusions: More time spent in education is causally associated with a large reduction in
the risk of coronary heart disease. This may be partly explained by changes to smoking, BMI
and a blood lipids. These findings offer support for policy interventions that increase
education, in order to also reduce the burden of cardiovascular disease.
Keywords: Socioeconomic Factors, Educational status, Cardiovascular disease, Myocardial
ischemia, Coronary artery disease, Myocardial infarction, Prospective studies, Mendelian
randomizaion
2
bioRxiv preprint first posted online Feb. 6, 2017; doi: http://dx.doi.org/10.1101/106237. The copyright holder for this preprint (which was not
peer-reviewed) is the author/funder. It is made available under a CC-BY 4.0 International license.
What this paper adds
What is already known on this subject: Numerous observational studies have found that
people with longer educational attainment develop less coronary heart disease. However it
is not clear whether this association is causal, partly since randomised controlled trials are
not feasible in this area. No prior studies have applied the Mendelian randomization
method to investigate how exposure to socioeconomic risk factors might causally change
the risk of disease occurrence.
What this study adds: Our study suggests that if young people were to increase the number
of years they spend in the educational system, then this is likely to substantially lower their
risk of subsequently developing coronary heart disease. Our study should stimulate further
policy discussion about increasing educational attainment in the general population, in
order to improve population health.
3
bioRxiv preprint first posted online Feb. 6, 2017; doi: http://dx.doi.org/10.1101/106237. The copyright holder for this preprint (which was not
peer-reviewed) is the author/funder. It is made available under a CC-BY 4.0 International license.
INTRODUCTION
Coronary heart disease (CHD) is the leading cause of death globally. While the causal effects
of risk factors like smoking, blood pressure and LDL-cholesterol are generally accepted and
reflected in disease prevention strategies, substantial uncertainty still surrounds other
potential factors. Decades of observational studies have consistently associated
socioeconomic factors such as higher education with decreased risk of CHD.1-4 However,
given the implausibility and consequent absence of randomized controlled trials in this field,
it is possible that this association does not stem from an underlying causal effect, but arises
due to the methodological flaws of observational research.5 6 Clarifying whether the
association between education and CHD is causal or not has widespread implications for our
understanding of the aetiology of CHD, and the development of novel population-based
approaches to its prevention.
Mendelian randomization analysis relies on genetic variants that are associated with a risk
factor (e.g. education), to make causal inferences about how environmental changes to the
same risk factor would alter the risk of disease (e.g. CHD).7 Genetic variants associated with
a risk factor are independent from confounders that may otherwise cause bias, as genetic
variants are randomly allocated before birth.8 This, as well as the non-modifiable nature of
the genetic variants, provides an analogy to trials, where exposure is allocated by random
and is non-modifiable by disease.8 By comparing the risk of disease across the genotype
groups, this allows causal effects to be determined with substantially less bias than that
from observational epidemiology. Recent methodological developments, including Egger
Mendelian randomization (MR-Egger), weighted median MR and multivariable MR, can be
employed as sensitivity analyses to additionally investigate any pleiotropic effects of the
genetic variants (i.e., when genetic variants associate with one or more pathways than with
4
bioRxiv preprint first posted online Feb. 6, 2017; doi: http://dx.doi.org/10.1101/106237. The copyright holder for this preprint (which was not
peer-reviewed) is the author/funder. It is made available under a CC-BY 4.0 International license.
the risk factor under analysis, which can distort the causal effect estimate) (Figure S1).9-11
Here, we first updated observational estimates of the association between education and
incidence of coronary heart disease from several large studies (Table S1). A recent genomewide association study (GWAS) from the Social Science Genetic Association Consortium
identified a large number of independent genetic variants (i.e., single-nucleotide
polymorphisms [SNPs]) associated with educational attainment.12 Thus, we used two sets of
SNPs associated with education attainment and retrieved the same CHD-associated SNPs
from the CARDIoGRAMplusC4D Consortium to perform Mendelian randomization analyses,
investigating whether individuals with a genetic predisposition towards higher education
have a lower risk of CHD.13 We checked the robustness of our findings across a range of
sensitivity analyses. We finally investigated whether genetic liability for CHD is associated
with educational outcomes.
METHODS
Observational association between education and CHD
Throughout all analyses, CHD was defined as a composite of myocardial infarction, acute
coronary syndrome, chronic stable angina or coronary stenosis of >50%, or coronary death.
In observational analysis, we used a combination of cross-sectional and prospective data,
collected between 1983-2014 (Table S1). For prevalent CHD cases, we analysed 43,611
participants (1,933 cases) from the National Health and Nutrition Examination Surveys
(NHANES) (Tables 1 and S1, Figure S2) (http://www.cdc.gov/nchs/nhanes).14 For incident
cases, we analysed 23,511 participants (632 CHD cases) from the Health, Alcohol and
5
bioRxiv preprint first posted online Feb. 6, 2017; doi: http://dx.doi.org/10.1101/106237. The copyright holder for this preprint (which was not
peer-reviewed) is the author/funder. It is made available under a CC-BY 4.0 International license.
Psychosocial factors In Eastern Europe (HAPIEE) study15 and combined this with published
estimates from 97,048 participants (6,522 cases) of the MOnica Risk, Genetics, Archiving
and Monograph (MORGAM) study in Europe (see Table S1 for case definitions, and Figures
S3-S4).16
Genetic variants associated with education
SNPs associated with educational attainment were retrieved from a recent genome-wide
association study, involving 405,072 individuals of European ancestry (Table 1).12 For our
main analysis, we used 162 independent SNPs (linkage disequilibrium, r2<0.1) associated
with education across the discovery and replication cohorts at a combined GWAS
significance threshold (P<5.10-8) (Supplementary Methods, Table S2-S3 and Figure S5-S6).
For our secondary analysis, we used another set of 72 independent SNPs (linkage
disequilibrium, r2<0.1) that were associated with education in the discovery cohort (293,723
participants; P<5.10-8) and directionally consistent in the replication cohort (Supplementary
Methods, Table S3-S4, and Figure S5 and S7).
Genetic variants associated with CHD
For each education-associated SNP, we retrieved summary-level data from the Coronary
ARtery DIsease Genome wide Replication and Meta-analysis plus The Coronary Artery
Disease Genetics Consortium (CARDIoGRAMplusC4D) comprising 63,746 CHD cases and
130,681 controls (Supplementary Methods, Table 1, and Figures S8-S9).13
Statistical analyses
Cox proportional hazards and logistic regressions were used to calculate observational
estimates for incident and prevalent cases, respectively. Conventional Mendelian
6
bioRxiv preprint first posted online Feb. 6, 2017; doi: http://dx.doi.org/10.1101/106237. The copyright holder for this preprint (which was not
peer-reviewed) is the author/funder. It is made available under a CC-BY 4.0 International license.
randomization analyses were performed by regressing the SNP-education associations with
the SNP-CHD associations (outcome), where each SNP was one data point (Supplementary
Methods).9
Sensitivity analyses (MR-Egger and weighted median MR) were used to investigate to what
degree pleiotropic effects might bias the Mendelian randomization causal estimates.9 10 In
addition, we also applied Mendelian randomization to investigate whether genetic
predisposition to higher education could lead to improvements in the established
cardiovascular risk factors (Supplementary Methods and Table S5). Once such associations
were identified, we further adjusted for their causal effects using multivariate MR.11 Finally,
to check for whether genetic risk for coronary events might be a causal factor for
educational attainment, we performed Mendelian randomization in the opposite direction
(Supplementary Methods and Table S6).
Patient involvement
Patients were not involved in the design or implementation of this study. There are no
specific plans to disseminate the research findings to participants, but findings will be
returned back to the original consortia, so that they can consider disseminating further.
RESULTS
Observational analyses
Based on NHANES data, each additional 3.6 years of education (1-SD) was associated with
27% lower odds of CHD (odds ratio, 0.73; 95% confidence interval [CI], 0.68 to 0.78) (Figure
7
bioRxiv preprint first posted online Feb. 6, 2017; doi: http://dx.doi.org/10.1101/106237. The copyright holder for this preprint (which was not
peer-reviewed) is the author/funder. It is made available under a CC-BY 4.0 International license.
2). In prospective analyses, 3.6 years of additional education was associated with a 20%
lower risk of CHD in the HAPIEE and MORGAM studies, with a pooled hazard ratio of 0.80
(95% CI, 0.76 to 0.83) (Figure S10).15 16 These observational estimates were robust to
sensitivity analyses accounting for different case definitions, different age at first event, and
confounding by other measures of socioeconomic position (Table S7).
Causal association of education on CHD
When examining SNPs across the entire genome (Supplementary Methods), there was
strong evidence for a negative genetic correlation between education and CHD (rg=-0.324;
p-value=2.1 x 10-12).17
Conventional Mendelian randomization analysis using 162 SNPs found that a 1-SD increase
in education was associated with a 33% reduction in the risk of CHD (odds ratio, 0.67; 95%
CI, 0.59 to 0.77) (Figure 1 and S11). Sensitivity analyses using MR-Egger and weighted
median Mendelian randomization yielded similar results in terms of direction and
magnitude (Figures 1 and S12-S13 and Table S8), suggesting that our findings were not
biased from presence of pleiotropic effects. The secondary analysis using a set of 72 SNPs
yielded consistent results in terms of direction and magnitude (Figures 1 and S14-S16, Table
S8).18
Causal association of education on cardiovascular risk factors
To identify potential risk factors that could mediate or confound the association between
education and CHD, we investigated whether a genetic predisposition towards higher
education was associated with established cardiovascular risk factors (Table 2 and S5). A 1SD increase in education was causally associated with a 35% reduction in odds of smoking, a
8
bioRxiv preprint first posted online Feb. 6, 2017; doi: http://dx.doi.org/10.1101/106237. The copyright holder for this preprint (which was not
peer-reviewed) is the author/funder. It is made available under a CC-BY 4.0 International license.
0.17 kg/m2 reduction in BMI, a 0.14 mmol/l reduction in triglycerides and a 0.15 mmol/l
increase in HDL-cholesterol (Table 2). In exploratory analyses, inclusion of genetic
associations with smoking, BMI, triglycerides and HDL-cholesterol did not change the
primary association between education and CHD (Tables 3 and S9).
Causal association of CHD on education
The reverse investigation, of whether genetic liabilities for risk of CHD are associated with
educational outcomes, showed an absence of association (Figures 2 and S17-S19).
DISCUSSION
In this Mendelian randomization study, we found evidence that higher education is causally
related to substantial reductions in the risk of CHD, consistent with observational estimates.
More specifically, 3.6 years of increased education (similar to an undergraduate university
degree) translates to about a one-third reduction in the risk of CHD.
Previous studies have attempted to clarify the causality of the relationship between
education and CHD using approaches that overcome some of the limitations of
observational epidemiology. These have primarily investigated the association between
education and all-cause mortality. Few have investigated cardiovascular outcomes, and
none have used a genetic approach using Mendelian randomization analysis. First,
ecological instrumental variable approaches have compared mortality before and after
changes to compulsory schooling laws. For example, by looking at mortality rates in
countries before and after the introduction of national legislation that increased minimum
education. In the UK, this increase in minimum education by one year was not associated
9
bioRxiv preprint first posted online Feb. 6, 2017; doi: http://dx.doi.org/10.1101/106237. The copyright holder for this preprint (which was not
peer-reviewed) is the author/funder. It is made available under a CC-BY 4.0 International license.
with a change in mortality, but a similar increase in the Netherlands was associated with
reduced mortality.5 19 Analyses of inter-state differences within USA initially suggested a
large effect on mortality, but this effect disappeared when state-specific baseline trends
were adjusted for.20 21 In Sweden, an intervention to extend compulsory schooling
throughout a 13-year transition period was not associated with lower mortality for all age
groups combined, but was associated with lower mortality in the younger age groups.22
Second, studies on twins have found differences between siblings in terms of development
of chronic diseases when accounting for education, suggesting that confounding from
environmental factors is unlikely to explain the observational associations.23-25 Third, studies
including individual-level data from millions of siblings observed an association between
education and mortality.6 26 Fourth, a recent study has reported a link between parental
longevity and genetic markers for education in their offspring.27 However, causation and its
direction were not tested.
The mechanisms that might mediate the association between education and CHD remain
relatively unknown. Observational associations have found that the primary association
attenuates by around 30-45% after statistical adjustment for health behaviours and
conventional cardiovascular risk factors (including smoking, blood pressure and cholesterol).
This suggests that these factors could account for around half of the association between
education and CHD.2 28 In this study, we observed that education is causally related to
smoking, BMI and blood lipids. Clarifying how much they mediate the causal association
between education and CHD requires further investigation, e.g. by applying two-step
Mendelian randomization to individual-level data.29 30 Nonetheless, accounting for these
factors in our multivariable Mendelian randomization analysis did not influence the primary
association between education and risk of CHD, suggesting that conventional risk factors
10
bioRxiv preprint first posted online Feb. 6, 2017; doi: http://dx.doi.org/10.1101/106237. The copyright holder for this preprint (which was not
peer-reviewed) is the author/funder. It is made available under a CC-BY 4.0 International license.
that we assessed may not completely account for the mechanism. Additional hypotheses for
investigation could include education leading to improved use of healthcare services (from
better health knowledge or fewer financial barriers in accessing care), better job prospects,
income, material conditions, social ranking and/or diet - all factors associated education and
CHD.4
Our study has important strengths. We investigated the causality of the association
between an easily measured socioeconomic factor (i.e. education) and a common disease
(i.e. coronary heart disease), using a genetic design to substantially reduce bias. Summarylevel data from over half a million individuals gave our study the precision required to derive
robust causal effect estimates, and to perform multiple sensitivity analyses. The consistency
of findings across two sets of genetic instruments strengthens the confidence in our
findings. We used recent state-of-the-art methodological developments to thoroughly
explore pleiotropy our genetic variants, for which we found no evidence. However, our
study also has some limitations. First, it is possible that the genetic variants associated with
education may instead mark more generic biological pathways (such as vascular supply or
mitochondrial function), which could simultaneously lead to increased educational
attainment and reduced risk of CHD.25 31 In this scenario, interventions in education may not
translate into lower disease occurrence. However, such a scenario is less likely to lead to the
consistent set of results we found across all the sensitivity analyses. Second, it is possible
that some of the genetic markers used were invalid. Nonetheless, even if all SNPs included
in the analysis have pleiotropic effects, MR-Egger analyses should still give valid causal
estimates, as long as the pleiotropic effects do not correlate with the effects on education.
Third, we assumed the absence of dynastic effects, an assumption that is broken when
parental genes associate with parental behaviours that directly cause a health outcome in
11
bioRxiv preprint first posted online Feb. 6, 2017; doi: http://dx.doi.org/10.1101/106237. The copyright holder for this preprint (which was not
peer-reviewed) is the author/funder. It is made available under a CC-BY 4.0 International license.
the child.32 For example, parents with a genetic predisposition towards higher education
may choose to feed their children a better diet. Fourth, our observational and genetic data
originate predominantly from European origin samples in high-income countries. We are
thus unable to generalize these estimates to other populations, particularly to low-income
countries where cardiovascular diseases are less common. Fifth, we do not know whether
increasing education for those of least education will be as cardioprotective as increasing
education for those with above-average education. Nonetheless, the linear relationship in
the observational data supports interventions framed within Rose’s prevention paradox:
that more disease could be prevented by slight increases in education across its entire
distribution, as opposed to targeting interventions only at high-risk subgroups.33 Sixth, we
assumed that genetic predisposition towards a higher educational attainment causes the
same behavioural and physiological consequences as environmentally-acquired changes to
educational attainment. Although our study did not allow us to directly test the non-genetic
component of education, these genetic findings are generally consistent with the various
studies observing differences in mortality following policy change to schooling laws (as
described above), as well as with observational estimates that measure the combined
genetic and acquired components of education. Moreover, genetic predispositions to e.g.
LDL-cholesterol and systolic blood pressure levels have recapitulated the direction of
causality seen in environmentally-acquired changes to these risk factors from randomized
controlled trials of pharmacological therapies, supporting the overall validity of the MR
approach.9 34
In conclusion, this study suggests that a causal relationship exists between more time spent
in education and a reduced risk of CHD. This relationship is only partly explained by
traditional cardiovascular risk factors. Our findings add to the growing evidence base that
12
bioRxiv preprint first posted online Feb. 6, 2017; doi: http://dx.doi.org/10.1101/106237. The copyright holder for this preprint (which was not
peer-reviewed) is the author/funder. It is made available under a CC-BY 4.0 International license.
increasing education is likely to lead to not only important societal benefits, but also
substantial health benefits.
Competing interests: All authors have completed the ICMJE uniform disclosure form at
www.icmje.org/coi_disclosure.pdf and declare: no financial relationships with any
organisations that might have an interest in the submitted work in the previous three years;
no other relationships or activities that could appear to have influenced the submitted
work.
Contributors: Idea: TT. Genetic Data: AO. Observational Data: HP, AP, RK, AP, AT, SM, KF.
Methods: MVH, GDS, JB, TP, GV. Analysis: JV, TT, MVH. Interpreted data: all co-authors.
Wrote first draft: TT, JV, MVH. Revised manuscript for critical revisions: all co-authors. TT,
JV and MVH are guarantors. TT, JV and MVH affirm that this manuscript is an honest,
accurate, and transparent account of the study being reported; that no important aspects of
the study have been omitted; and that any discrepancies from the study as planned have
been explained.
Acknowledgements: TT is funded by a Wellcome Trust Fellowship [106554/Z/14/Z]; JV is
supported by the Swiss National Science Foundation (P2LAP3_155086); The HAPIEE study is
supported by the Wellcome Trust [064947/Z/01/Z, WT081081]; the US National Institute on
Aging [1R01 AG23522]; the MacArthur Foundation [Health and Social Upheaval network];
and the Russian Science Foundation [14-45-00030]; GDS works in the Medical Research
Council Integrative Epidemiology Unit at the University of Bristol [MC_UU_12013/1]. The
funders, had no role in the study design, data collection, analysis, interpretation, writing,
nor the decision to submit the article for publication.
13
bioRxiv preprint first posted online Feb. 6, 2017; doi: http://dx.doi.org/10.1101/106237. The copyright holder for this preprint (which was not
peer-reviewed) is the author/funder. It is made available under a CC-BY 4.0 International license.
We are grateful to two large consortia (SSGAC and CARDIoGRAMplusC4D) for publically
sharing the genetic data we used in our causal analysis.
REFERENCES
1.
Hinkle LE, Jr., Whitney LH, Lehman EW, et al. Occupation, education, and coronary heart
disease. Risk is influenced more by education and background than by occupational
experiences, in the Bell System.
Science 1968;16:238-46.
2.
Manrique-Garcia E, Sidorchuk A, Hallqvist J, et al. Socioeconomic position and incidence of
3.
Veronesi G, Ferrario MM, Kuulasmaa K, et al. Educational class inequalities in the incidence of
4.
Davey Smith G, Hart C, Hole D, et al. Education and occupational social class: which is the more
5.
Clark D, Royer H. The effect of education on adult mortality and health: evidence from Britain.
acute myocardial infarction: a meta-analysis.
coronary heart disease in Europe.
J Epidemiol Community Health 2011;65:301-9.
Heart 2016;102:958-65.
important indicator of mortality risk?
J Epidemiol Community Health 1998;52:153-60.
Am Econ Rev 2013;103:2087-120.
6.
Naess O, Hoff DA, Lawlor D, et al. Education and adult cause-specific mortality--examining the
7.
Davey Smith G, Ebrahim S. 'Mendelian randomization': can genetic epidemiology contribute to
8.
Davey Smith G, Ebrahim S. What can mendelian randomisation tell us about modifiable
impact of family factors shared by 871 367 Norwegian siblings.
Int J Epidemiol 2003;32:1-22.
understanding environmental determinants of disease?
behavioural and environmental exposures?
Int J Epidemiol 2012;41:1683-91.
BMJ 2005;330:1076-9.
9.
Bowden J, Davey Smith G, Burgess S. Mendelian randomization with invalid instruments: effect
10.
Bowden J, Davey Smith G, Haycock PC, et al. Consistent Estimation in Mendelian Randomization
estimation and bias detection through Egger regression.
Int J Epidemiol 2015;44:512-25.
with Some Invalid Instruments Using a Weighted Median Estimator.
Genet Epidemiol
2016;40:304-14.
11.
Burgess S, Dudbridge F, Thompson SG. Re: "Multivariable Mendelian randomization: the use of
pleiotropic genetic variants to estimate causal effects".
12.
Am J Epidemiol 2015;181:290-91.
Okbay A, Beauchamp JP, Fontana MA, et al. Genome-wide association study identifies 74 loci
associated with educational attainment.
Nature 2016;533:539-42.
13.
Deloukas P, Kanoni S, Willenborg C, et al. Large-scale association analysis identifies new risk loci
14.
Centers for Disease Control and Prevention (CDC). National Center for Health Statistics (NCHS).
for coronary artery disease.
Nat Genet 2013;45:25-33.
National Health and Nutrition Examination Survey Data. Hyattsville, MD: U.S. Department of
Health and Human Services, Centers for Disease Control and Prevention, 1999-2014.
http://www.cdc.gov/nchs/nhanes/, Accessed 2016, Aug 27.
15.
Peasey A, Bobak M, Kubinova R, et al. Determinants of cardiovascular disease and other noncommunicable diseases in Central and Eastern Europe: rationale and design of the HAPIEE
study. BMC public health 2006;6:255. doi: 10.1186/1471-2458-6-255.
16.
Evans A, Salomaa V, Kulathinal S, et al. MORGAM (an international pooling of cardiovascular
cohorts).
17.
Int J Epidemiol 2005;34:21-7.
Zheng J, Erzurumluoglu AM, Elsworth BL, et al. LD Hub: a centralized database and web
interface to perform LD score regression that maximizes the potential of summary level GWAS
data for SNP heritability and genetic correlation analysis.
Bioinformatics 2016. doi:
10.1093/bioinformatics/btw613. Epub ahead of print.
14
bioRxiv preprint first posted online Feb. 6, 2017; doi: http://dx.doi.org/10.1101/106237. The copyright holder for this preprint (which was not
peer-reviewed) is the author/funder. It is made available under a CC-BY 4.0 International license.
18.
Bowden J, Del Greco MF, Minelli C, et al. Assessing the suitability of summary data for twosample Mendelian randomization analyses using MR-Egger regression: the role of the I2
statistic.
19.
Int J Epidemiol 2016. doi: 10.1093/ije/dyw220. Epub ahead of print.
Baigi A, Holmen A, Hogstedt B, et al. Birthplace and social characteristics as risk factors for acute
myocardial infarction in the province of Halland, Sweden. Public health 2002;116(5):279-84.
20.
Lleras-Muney A. The relationship between education and adult mortality in the United States.
Rev Econ Stud 2005;72:189-221.
21.
Mazumder B. Does education improve health? A reexamniation of the evidence from
22.
Lager AC, Torssander J. Causal effect of education on mortality in a quasi-experiment on 1.2
compulsory schooling laws.
million Swedes.
23.
Econ Perspect 2008;33:2-16.
Proc. Natl. Acad. Sci. USA 2012;109:8461-6.
Lundborg P. The health returns to schooling - what can we learn from twins?
J Pop Econ
2013;26:673-701.
24.
Madsen M, Andersen AM, Christensen K, et al. Does educational status impact adult mortality
in Denmark? A twin approach.
Am J Epidemiol 2010;172:225-34.
25.
Arden R, Luciano M, Deary IJ, et al. The association between intelligence and lifespan is mostly
26.
Sondergaard G, Mortensen LH, Nybo Andersen AM, et al. Does shared family background
genetic.
Int J Epidemiol 2016;45:178-85.
influence the impact of educational differences on early mortality?
Am J Epidemiol
2012;176:675-83.
27.
Marioni RE, Ritchie SJ, Joshi PK, et al. Genetic variants linked to education predict longevity.
28.
Stringhini S, Sabia S, Shipley M, et al. Association of socioeconomic position with health
29.
Richmond RC, Hemani G, Tilling K, et al. Challenges and novel approaches for investigating
30.
Varbo A, Benn M, Smith GD, et al. Remnant cholesterol, low-density lipoprotein cholesterol, and
31.
Nelson CP, Hamby SE, Saleheen D, et al. Genetically determined height and coronary artery
32.
Fletcher JM. The promise and pitfalls of combining genetic and economic research.
Proc. Natl. Acad. Sci. USA 2016;113:13366-13371.
behaviors and mortality.
molecular mediation.
JAMA 2010;303:1159-66.
Hum Mol Genet 2016;25:R149-R56.
blood pressure as mediators from obesity to ischemic heart disease.
disease.
Circ Res 2015;116:665-73.
N Engl J Med 2015;372:1608-18.
Health Econ
2011;20:889-92.
Int J Epidemiol 1985;14:32-8.
33.
Rose G. Sick individuals and sick populations.
34.
Swerdlow DI, Preiss D, Kuchenbaecker KB, et al. HMG-coenzyme A reductase inhibition, type 2
diabetes, and bodyweight: evidence from genetic analysis and randomised trials.
Lancet
2015;385:351-61.
15
bioRxiv preprint first posted online Feb. 6, 2017; doi: http://dx.doi.org/10.1101/106237. The copyright holder for this preprint (which was not
peer-reviewed) is the author/funder. It is made available under a CC-BY 4.0 International license.
TABLES
Table 1 Details of the studies and datasets included in the analysis
Analysis/study
Risk factor/Outcome
No. of
Web source (if publicly available)
participants
(CHD cases)
Observational analysis
Years of
43,611
NHANES
education/nonfatal
http://www.cdc.gov/nchs/nhanes
(1,933)
CHD
Years of
HAPIEE
15
education/fatal and
23,511 (632)
-
nonfatal CHD
Years of
MORGAM
16
97,048
education/fatal and
(6,522)
nonfatal CHD
Mendelian randomization analysis
SSGAC
12
CARDIoGRAMplusC4D
Years of education
13
349,306 (-)
http://www.thessgac.org/#!data/kuzq8
194,427
nonfatal CHD
http://www.cardiogramplusc4d.org
(63,746)
CHD, coronary heart disease. NHANES, The National Health and Nutrition Examination Survey. HAPIEE, Health,
Alcohol and Psychosocial factors In Eastern Europe. MORGAM, MOnica Risk, Genetics, Archiving and
Monograph. SSGAC, Social Science Genetic Association Consortium. CARDIoGRAMplusC4D, Coronary ARtery
DIsease Genome wide Replication and Meta-analysis (CARDIoGRAM) plus The Coronary Artery Disease (C4D)
Genetics) consortium
16
bioRxiv preprint first posted online Feb. 6, 2017; doi: http://dx.doi.org/10.1101/106237. The copyright holder for this preprint (which was not
peer-reviewed) is the author/funder. It is made available under a CC-BY 4.0 International license.
Table 2 Causal effects of a 1-SD increase in education on selected cardiovascular risk factors
Estimate (95% CI) per
1-SD increase in
education (~3.6 years)
Outcome
Binary traits
P-value
(estimate is Odds Ratio)
Smoking status
0.65 (0.54; 0.79)
0.001
Diabetes mellitus, type 2
0.75 (0.56; 1.01)
0.057
Continuous traits (units)
(estimate is beta)
Systolic Blood Pressure (mm Hg)
-1.36 (-2.85; 0.12)
0.075
Diastolic Blood Pressure (mm Hg)
-0.23 (-1.22; 0.76)
0.645
LDL-cholesterol (mmol/L)
-0.03 (-0.10; 0.05)
0.513
0.15 (0.07; 0.23)
0.001
-0.14 (-0.22; -0.06)
0.001
-0.02 (-0.08; 0.03)
0.441
-0.17 (-0.26; -0.08)
0.001
0.06 (-0.03; 0.16)
0.208
HDL-cholesterol
Triglycerides
(mmol/L)
(mmol/L)
Glucose (mmol/L)
Body Mass Index
2
(kg/m )
Height (cm)
All analyses are based on a common set of 102 single nucleotide polymorphisms associated with education,
measured in various international consortia (See
Supplementary Methods and Table S5). Bold font denotes
causal effect estimates with strong statistical evidence (p<0.001). Estimates are expressed per 1-standard
deviation increase in years of education (equivalent to 3.6 years) as absolute values for continuous risk factors,
and as odds ratios for binary traits. CI, confidence interval. LDL, low density lipoprotein; HDL, high density
lipoprotein.
17
bioRxiv preprint first posted online Feb. 6, 2017; doi: http://dx.doi.org/10.1101/106237. The copyright holder for this preprint (which was not
peer-reviewed) is the author/funder. It is made available under a CC-BY 4.0 International license.
Multivariable Mendelian randomization analyses showing the causal effect of 1-SD
increase in education on the risk of CHD
Table 3
Odds Ratio of CHD
(95% CI) per 1-SD
increase in
education (~3.6
years)
P-value
Attenuation
Nil (unadjusted)
0.79 (0.66; 0.94)
0.010
reference
Tobacco (ever vs. never users)
0.80 (0.66; 0.97)
0.023
Body Mass Index
0.81 (0.67; 0.98)
0.035
Triglycerides
0.76 (0.63; 0.92)
0.006
HDL-cholesterol
0.81 (0.67; 0.98)
0.033
All 4 risk factors
0.80 (0.65; 0.99)
0.039
Risk factor adjusted for
5%
11%
-16%
11%
5%
All analyses are based on a common set of 102 single nucleotide polymorphisms associated with education,
measured in 63 746 CHD cases and 130 681 controls, in the CARDIoGRAMplusC4D Consortium (See
Supplementary Methods and Table S5). SD, standard deviation. CHD, coronary heart disease. CI, confidence
interval HDL, high density lipoprotein.
18
bioRxiv preprint first posted online Feb. 6, 2017; doi: http://dx.doi.org/10.1101/106237. The copyright holder for this preprint (which was not
peer-reviewed) is the author/funder. It is made available under a CC-BY 4.0 International license.
FIGURES
Comparison of observational (blue) and causal (red) estimates for risk of coronary
heart disease, per 3.6 years of educational attainment
Figure 1
Two observational estimates (in blue) are provided according to incident or prevalent CHD cases. The
incidence risk coefficient (RC) was derived by meta-analysis of hazard ratios from the HAPIEE and MORGAM
studies. The observational estimates are contrasted with the different Mendelian randomization (MR) causal
effect estimates (in red) (see Supplementary Methods for a full description of each MR approach). The RCs for
the observational prevalence estimate and the causal estimates are odds ratios. CHD, coronary heart disease.
CI, Confidence Interval. NHANES, The National Health and Nutrition Examination Survey. HAPIEE, Health,
Alcohol and Psychosocial factors In Eastern Europe. MORGAM, MOnica Risk, Genetics, Archiving and
Monograph. IVW, Inverse-variance weighted approach.
19
bioRxiv preprint first posted online Feb. 6, 2017; doi: http://dx.doi.org/10.1101/106237. The copyright holder for this preprint (which was not
peer-reviewed) is the author/funder. It is made available under a CC-BY 4.0 International license.
Figure 2 Association of genetic liabilities for risk of CHD and educational outcomes
Causal estimates are expressed as SD difference in education, per 1-log unit increase in risk of coronary heart
disease (CHD). Each Mendelian randomization (MR) analysis is fully described in the Supplementary Methods.
CI, confidence interval. IVW, inverse variance weighted approach.